Hybrid LQG-Neural Controller for Inverted Pendulum System

Title: Hybrid LQG-Neural Controller for Inverted Pendulum System

Research Question: How can a hybrid system controller, incorporating a neural network and an LQG controller, be designed to optimize the regulation process and maintain stability for an inherently unstable system like an inverted pendulum?

Methodology: The researchers designed a hybrid controller consisting of a neural controller and an LQG controller. The neural controller was optimized using genetic algorithms to ensure a failure-free optimization process. The LQG controller was designed to provide stability during transient processes and a wide range of operation. Both controllers were integrated into a single system controller to leverage the benefits of each.

Results: The hybrid controller was successfully applied to a simulation model of an inverted pendulum, demonstrating high-quality regulation and stability. The neural controller ensured a high-quality regulation process, while the LQG controller provided stability during transient processes and a wide range of operation.

Implications: The hybrid controller presented in this research offers a solution to the challenge of designing effective controllers for inherently unstable systems. By combining the benefits of a genetically optimized neural controller and an LQG controller, it ensures both high-quality regulation and stability. This approach can be applied to other complex systems where traditional control methods may not be sufficient.

Link to Article: https://arxiv.org/abs/0312003v1 Authors: arXiv ID: 0312003v1